Inspiration

Endless scrolling is often treated as a simple problem of self-control. But for people experiencing attention regulation difficulties, boredom, emotional avoidance, or challenges switching between tasks, disengaging from digital content can be much more complicated.

I wanted to explore a different approach: instead of punishing users for scrolling or simply blocking their favorite apps, what if technology could help them understand their behavior and make more intentional choices?

That idea led to Unhook: Escape the Loop, an interactive, AI-assisted behavioral wellness application that combines psychology-informed reflection, gamification, and small real-world actions.

What it does

Unhook is a mobile-friendly progressive web app designed to help users interrupt unwanted scrolling habits through short, engaging experiences.

The application includes:

  • Nine interactive skill games: Playable challenges inspired by cognitive behavioral therapy (CBT) concepts, targeting procrastination, impulsivity, boredom, and reward-seeking.
  • AI Behavioral Coach: Optional personalized reflective questions before and after gameplay.
  • Smart recommendations: AI-assisted suggestions for games and small actions based on users' check-ins, with a rule-based alternative.
  • Psychology-inspired learning: Five micro-lessons that introduce relevant behavioral concepts.
  • Personal Journey: Locally saved progress, reflections, and insights.
  • ReelWorld: A simulated social media environment demonstrating how an intervention could appear before scrolling.

Rather than focusing exclusively on reducing screen time, Unhook encourages users to recognize their triggers, reflect on their needs, and experiment with alternative behaviors.

How I built it

Unhook was developed as a progressive web application using JavaScript, HTML, CSS, Node.js, and Express.

The frontend handles interactive games, check-ins, progress tracking, and navigation. The backend integrates the OpenAI Responses API to generate optional contextual recommendations and behavioral reflections.

The architecture combines AI-generated responses with deterministic, rule-based recommendations. This allows the core experience to function even when an AI API key is unavailable.

User progress is stored locally using IndexedDB, with a localStorage fallback. Users can export, import, and erase their data.

Challenges

An important design challenge was balancing psychological relevance with an engaging, accessible experience.

Behavioral reflection can require considerable effort, particularly for users already struggling with motivation or task initiation. Unhook addresses this by integrating short check-ins into interactive games rather than relying entirely on lengthy questionnaires.

Another challenge was designing an AI-assisted experience that remains useful without continuous AI access, while keeping model-generated feedback within appropriate educational boundaries.

Accomplishments

The prototype combines nine playable games, optional AI coaching, a recommendation system, short educational content, and local progress tracking within one application.

A particularly important achievement is integrating psychological concepts into interactive gameplay while providing a non-AI fallback and user-controlled data storage.

What I learned

Developing Unhook highlighted the importance of connecting behavioral concepts with practical interaction design.

AI can support contextual reflection, but structured interactions, privacy-conscious architecture, and transparent limitations are equally important.

What's next

Future development could include usability testing, accessibility improvements, additional behavioral games, and research into how users engage with brief interventions.

A longer-term goal is to investigate whether these experiences can support intentional technology use through appropriately designed user studies.

Unhook is an educational behavioral-wellness prototype, not a clinically validated treatment, diagnostic tool, or medical device.

Built With

Share this project:

Updates

Submission history